Anthropic Expands Project Glasswing: Claude Mythos Preview Now Accessible to 150 Organizations
Anthropic has significantly broadened the scope of Project Glasswing, its collaborative AI-driven cybersecurity initiative, by extending access to the Claude Mythos Preview to approximately 150 additional organizations. This expansion follows weeks of close collaboration with existing partners, the broader security industry, open-source software maintainers, and the U.S. government.
Project Glasswing’s Evolution
Initially launched in early April 2026, Project Glasswing began with around 50 partner organizations utilizing the Claude Mythos Preview to scan their codebases for security vulnerabilities. These early adopters collectively identified over 10,000 high- or critical-severity security flaws, highlighting the potent detection capabilities of Mythos-class AI models when applied to enterprise and infrastructure codebases.
Diverse Global Participation
The newly onboarded organizations span more than fifteen countries, with many providing critical infrastructure services that extend beyond their national borders. Industries previously underrepresented, such as power, water, healthcare, communications, and hardware, are now integrated into the program. A significant portion of these new partners are vendors and nonprofits maintaining widely used codebases relied upon by organizations worldwide, including national governments.
Anthropic’s risk assessment is clear: for most Project Glasswing partners, a successful cyberattack on their codebase could impact over 100 million people, posing significant threats to both global and national security. Each new organization must meet Anthropic’s stringent security requirements before gaining access to the models.
Enhanced Capabilities of Claude Mythos Preview
Beyond raw vulnerability detection, partners are now deploying the Claude Mythos Preview across a broader range of defensive tasks. These include:
– Automated patch writing
– Pre-release security checks to prevent vulnerabilities from entering production
– Penetration testing simulations
– Threat detection and response automation
– Rebuilding legacy codebases in memory-safe languages
The Claude Mythos Preview also serves as a critical warning signal for the industry. Anthropic estimates that within six to twelve months, competing AI companies will develop Mythos-class models, potentially releasing them without sufficient safeguards to prevent offensive misuse. This timeline creates an urgent window for cyber defenders to adapt their tools and operational norms before the threat landscape shifts dramatically.
Scaling Defensive Capabilities
To help scale these defensive capabilities beyond Project Glasswing’s direct partners, Anthropic is releasing, upon request to trusted security teams, the tools developed to support the program’s vulnerability detection workflows. The company has also launched Claude Security, a product built on its frontier public models, including Claude Opus 4.8, designed to scan codebases and suggest patches for organizations outside the program.
Anthropic is actively engaging with third parties to substantially scale up the review and patching of vulnerabilities in open-source software. This includes developing standardized best practices for disclosing findings to open-source maintainers to streamline triage and remediation processes.
Looking Ahead
Anthropic acknowledges that hundreds of thousands of organizations will ultimately need access to Mythos-level capabilities to address the evolving cybersecurity landscape. The company is committed to expanding access responsibly, ensuring that organizations can leverage advanced AI tools to bolster their security postures effectively.
